petermartens98/Qwen3-LLM-Pytorch-Implementation-From-Scratch

Lightweight LLM inspired by Qwen3, built from scratch in PyTorch. Full training pipeline with transformer components including RMSNorm, Rotary Position Embeddings (RoPE), Grouped-Query Attention (GQA), and SwiGLU layers. Trained with hybrid Muon + AdamW optimizer, causal masking, efficient batching, and evaluation tools.

22
/ 100
Experimental
No License No Package No Dependents
Maintenance 6 / 25
Adoption 3 / 25
Maturity 1 / 25
Community 12 / 25

How are scores calculated?

Stars

3

Forks

1

Language

Jupyter Notebook

License

Last pushed

Oct 20, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/petermartens98/Qwen3-LLM-Pytorch-Implementation-From-Scratch"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.